Market design for system security in low-carbon electricity grids: from the physics to the economics
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
June 2020 Market design for system security in low-carbon electricity grids: from the physics to the economics Farhad Billimoria, Pierluigi Mancarella and OIES Paper: EL 41 Rahmatallah Poudineh
The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its members. Copyright © 2020 Oxford Institute for Energy Studies (Registered Charity, No. 286084) This publication may be reproduced in part for educational or non-profit purposes without special permission from the copyright holder, provided acknowledgment of the source is made. No use of this publication may be made for resale or for any other commercial purpose whatsoever without prior permission in writing from the Oxford Institute for Energy Studies. ISBN 978-1-78467-160-0 DOI: https://doi.org/10.26889/9781784671600 i
Market design for system security in low-carbon electricity grids: from the physics to the economics Farhad Billimoria Visiting Research Fellow, Oxford Institute for Energy Studies Pierluigi Mancarella The University of Melbourne and The University of Manchester Rahmatallah Poudineh Senior Research Fellow and Director of Research, Oxford Institute for Energy Studies Abstract System security is a critical component of power system operation. The objective of operational security is to manage grid stability and to limit the interruption to customer service following a disturbance. The integration of inverter-based renewable generation technologies such as solar and wind in the generation mix has introduced new challenges for managing operational security. First, the intermittency inherent in renewable resources can impact on key power system parameters such as frequency and voltage. Second, renewables interface with the grid through power electronics rather than turbines, which means that the physical characteristics of turbine generation that have historically supported the stability of the grid are becoming scarce as the power system transitions away from fossil-fuel based thermal generation. In this paper, using public good theory, we provide an economic characterisation of the system services necessary for power system security. The analysis illustrates that, as opposed to a standard ‘public goods’ characterisation, system security products are better viewed as a basket of goods with differing economic characteristics that can also vary over space and time. The implications of these classifications for market design is analysed, including the inseparability of certain products, the binary or unit-commitment based nature of certain products, and the interactions between procurement mechanisms, network access rights and investment. Finally, we highlight five emerging models of market design for system security that reflect the nuances of economic characterization while respecting the physical characteristics of the grid. ii
Contents Abstract ................................................................................................................................................... ii Figures ................................................................................................................................................... iii Tables..................................................................................................................................................... iv 1. Introduction ......................................................................................................................................... 5 2. Emerging system security issues in the energy transition .................................................................. 6 2.1 An overview of power system security and stability ...................................................................... 6 2.2 Components of power system stability ......................................................................................... 7 2.3 Technology transition and integration ........................................................................................... 9 2.4 Power system requirements ....................................................................................................... 11 3. The economic characterization of system security ........................................................................... 15 3.1 The system services value chain: elementary services, system products and power system requirements ..................................................................................................................................... 15 3.2 An economic characterization of system security and services ................................................. 18 3.3 Developments in regulation and market design .......................................................................... 24 3.4 Implications of classification for market design ........................................................................... 25 4. Emerging design models for operational security ............................................................................. 27 4.1 Comprehensive central procurement (CP model) ...................................................................... 28 4.2 Decentralized procurement via cost or quantity allocation (DP-CQ model) ............................... 30 4.3 Decentralized access-driven procurement (DP-A model) ........................................................... 32 4.4 Hybrid arrangements................................................................................................................... 34 5. Summary and conclusions ................................................................................................................ 38 6. Appendix A: Examples illustrating the rivalry properties of system security services ...................... 41 6.1 Rivalry properties: Inertia and frequency response .................................................................... 41 6.2 Rivalry properties of fault levels (system strength) ..................................................................... 43 6.3 Rivalry properties of voltage control ............................................................................................ 45 Bibliography .......................................................................................................................................... 48 Figures Figure 1: Phase angle ............................................................................................................................. 8 Figure 2: Grid-following interface ............................................................................................................ 9 Figure 3: Frequency response under contingency ............................................................................... 12 Figure 4: Constance frequency operating modes ................................................................................. 13 Figure 5: Voltage stability margins ........................................................................................................ 14 Figure 6: Effects of reactive power on voltage ...................................................................................... 14 Figure 7: System security value chain – elementary services, system security products and requirements ......................................................................................................................................... 16 Figure 8: Voltage stability margins ........................................................................................................ 22 Figure 9: Depletion of fault levels through the addition of new inverter based generator .................... 23 Figure 10: Spectrum of classifications of system security products ..................................................... 24 Figure 11: Market and non-market mechanisms for system services .................................................. 25 Figure 12: Spectrum of access regimes ............................................................................................... 26 Figure 13: Central procurement (CP model) ......................................................................................... 29 Figure 14: Decentralized procurement via cost or quantity allocation (DP-CQ model) ........................ 31 Figure 15: Decentralized access-driven procurement (DP-A model) ................................................... 32 iii
Figure 16: Hybrid decentralized/centralized procurement (HD-C model) ............................................. 34 Figure 17: Decentralized procurement with facilitated co-ordination (HD-FC model) .......................... 36 Figure 18: Frequency response - swing equation ................................................................................. 41 Figure 19: Frequency parameters under contingency .......................................................................... 42 Figure 20: The impact on fault levels of a new user connecting to a network ...................................... 44 Figure 21: Simple two bus example – static voltage stability ................................................................ 45 Figure 22: Load voltage as a function of load real power ..................................................................... 46 Figure 23: Impact of increased reactive power on voltage stability curves .......................................... 47 Tables Table 1: Summary of technical power system requirements ................................................................ 11 Table 2: Classification criteria for different types of goods ................................................................... 18 Table 3: Summary of system service procurement models .................................................................. 28 Table 4: Advantages and disadvantages of system service procurement models ............................... 39 iv
1. Introduction A technological revolution is underway in most electricity systems around the world. Solar and wind generation technologies offer emission free energy at close to zero marginal cost. Coupling these technologies with storage and demand response allows for firmer sources of power and improved balance between power demand and supply. However, the integration of these technologies into the grid comes with new challenges for maintaining operational security. 1 Operational security relates to the ability of the power system to remain stable in the event of disturbances and keeping a system operationally secure is an everyday (and an every-minute) challenge. Recent events have underscored the criticality of system security – including the South Australia blackout, fault-driven disturbances in California, a regional blackout in the UK, and ongoing regional electrical separations within the National Electricity Market in Australia (AEMO, 2020a; National Grid, 2020a; NERC, 2019). Renewable technologies have introduced two major issues of concerns for operational security. First, is the variability and uncertainty inherent in resources such as wind and solar, which can impact upon power system parameters such as frequency and voltage. The second concern relates to the way in which renewables interface with the grid – through power electronics rather than turbines. There are physical characteristics of turbine generation interfaced via synchronous machines that have historically supported the stability of the grid – such as inertia2 and fault levels.3 With the phase-out of synchronous generation, grids around the world are experiencing challenges to maintain operational security. To manage security when critical system services are scarce, operators may need to resort to interventions in the market, renewable curtailment, and potentially even delaying new connections to the network. This has the potential to slow down the investment in renewables, which is critical to the deep decarbonization of the electricity sector. On the other hand, new technology solutions are being developed that use the speed, precision and control of power electronics to better manage security issues. However, without appropriate policy and market frameworks there is little incentive for market participants to deploy these advanced functionalities. System security is a multi-faceted concept and concerns many different aspects of power system physics. The challenge for policy makers and market designers is to develop cohesive economic and regulatory frameworks for operational security – and to manage integration within the broader market design for electric supply. An array of instruments is available to the policy maker, from regulation, mandatory licenses and markets (both organized and informal). An understanding of the economic characteristics of system security can inform the choice of which instrument or combination thereof is best suited for the security problem. We propose a bottom-up approach that begins with an understanding of the physics of electricity networks and translate this to an economic characterization. Indeed, it is these approaches that were the basis of foundational electricity market design in the first place (Joskow and Schmalensee, 1988; Schweppe et al., 1988). However, at that time markets for 1 In this paper, we are mainly concerned with the challenge of operational security. There are also challenges relating to resource adequacy and system balancing under deep power system decarbonisation. While we will note linkages with these challenges, a review of the optimal market designs for latter is out of the scope of this paper. In general we will assume that the high level market design of optimal least-cost generator dispatch based on short run marginal cost (SRMC) pricing principles is taken as a given at least from the perspective of short term resource commitment, scheduling and dispatch. 2 Mechanical energy stored in the rotating mass of the turbine which is released in the event of a disturbance. The primary effect of inertia is to resist the degradation of frequency, but it also affects transient and voltage stability. 3 Fault levels are an indicator of the maximum current that would flow to the fault at a specific node on the occurrence of a fault. There are often three sources of fault current: (i) large power plants via networks (i.e. system-derived fault current); (ii) embedded generators connected to the local network; and (iii) conversion of the mechanical inertia of the rotating turbine into electrical energy. Fault current is important to supporting power system stability in response to a fault. 5 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
energy and ancillary services were able to be designed around the assumption that connecting resources were primarily synchronous, as this was the predominant technology interface. With that assumption now challenged (and some would argue broken) frameworks for system security will need to adapt to the new emerging technology environment. The novel contributions of this paper are as follows: (1) a cohesive framing of the system services value chain, from elementary services to system security products to power system security requirements; (2) an application of public good theory to a variety of system security products to understand their economic characteristics; and (3) the application of this characterisation to five emerging economic frameworks for security. In doing so we illustrate that the economic characterization of system security products is increasingly heterogenous. Indeed, rather than thinking of all system security products as pure public goods, they can be better thought of as a basket of goods with public and private qualities. This necessitates a nuanced approach to regulatory and market design that reflects the interplay between network access, resource control and system characteristics. The paper is structured under three main parts. In Section 2 we will first outline the interactions between mechanics and electricity that have underpinned power system security to date. Then, we will highlight how the technology transition has fundamentally shifted the pillars of control that have underpinned power system security to date. In Section 3 we will begin the economic characterization by describing the system services value chain – from the operational elements and differing modes of control that go into creating different power system security products, to how they contribute to satisfying power system requirements. We then apply the perspective of public good theory to characterize these system security products. Next, armed with this lens, in Section 4 we set out five models for regulatory and market design and provide suitable pathways for policy in grids with differing technical characteristics and topologies. Finally, we provide some concluding remarks in Section 5. 2. Emerging system security issues in the energy transition 2.1 An overview of power system security and stability The security of the power system refers to its ability to remain stable in response to disturbances. Thus, the security is intrinsically linked with the concept of power systems engineering concept of stability. Stability is important across many dimensions – temporally, spatially and across different parameters of the system (such as frequency, phase angle, voltage, active and reactive power). Any imbalance in one dimension or element of the power system can quickly affect others. In reality, the system and its resources are built in a way that means it is able tolerate some level of imbalance, as long as this is arrested, mitigated and rebalanced within a sufficiently short period of time. Events such as generator or load trips, transmission line outages or unpredicted changes in renewable generation happen as a matter of course. From time to time extreme events such as multiple disconnections, extreme weather and cascading failures can also occur. At a high level, the goal of operational security is to manage those imbalances to limit the interruption to customer service (Kirschen and Strbac, 2004; Stoft, 2002). This involves handling the issue on multiple levels. First, to limit the risk and magnitude of those imbalances to within the operational tolerances of the system. Second, to ensure that there are sufficient emergency mechanisms to mitigate impacts when the tolerances are breached. Third, to recover the system effectively and expeditiously when there is an interruption to service. Finally, to ensure that the inevitable interactions between different aspects of security are manageable. 4 4 While this paper only focuses on the aspects of system security that are relevant for economic characterization, there are excellent technical treatments of the issue. Interested readers are referred to fundamental texts such as Glover et al. (2010), Kundur (1994) and for technical analyses - Kroposki et al., (2017); Milano et al., (2018); Perez-Arriago and Batlle (2012) and CIGRE (2016). 6 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
The power system parameters, such as frequency and voltage, are important indicators of the aforementioned stability in the system. To date, much of the focus of operational security has been upon managing these parameters and on keeping the system resilient to disturbances that affect them. The disturbance events that affect these parameters can be exogenous or endogenous, in other words, they can be triggered by an event that is external to the power system (such as weather events) or by elements within the power system itself (such as a trip of a generation unit). Moreover, it is often not just one single cause that results in the system deviating from its balance state – one event can impact another element of the power system, which affects another, and so on (i.e. a cascading failure). Increasingly, power system parameters can themselves interact with each other with adverse impacts. Take the case of ‘voltage dip-induced frequency dips’ or VDIFD – a phenomenon where a short temporary drop in the voltage magnitude induces a similar effect in the system frequency, which is observed more in systems with high renewable penetration (Rather and Flynn, 2015). In power systems with significant renewables, low voltage events can cause inverter-based resources (IBRs)5 to enter into an emergency control scheme (low voltage ride through) which limits their active power output for a short period of time. This loss of power can result in a frequency deviation, which must be managed in addition to the recovery of voltage. Indeed, analyses of system blackouts illustrate that blackouts are often not triggered by a single cause or impact, but by multiple interactive factors that ultimately result in interruption to service (Dobson, 2016; Pourbeik et al., 2006; Yamashita et al., 2009). These interactional factors will have increased relevance in the power systems of tomorrow given the mix of different technologies and sources of supply. Monitoring and managing these interactions are essential to having a secure and resilient power system. The goal of regulation and policy, in this context, is to develop mechanisms for the efficient management of power system security risks. This may involve both market instruments (such as spot markets or contracts) and non-market mechanisms (licenses, regulation). 2.2 Components of power system stability The physics of legacy AC power systems rely upon interactions between the physics of mechanical systems of generators and the physics of electrical networks; for example, the mechanics required to keep all turbines spinning at the same speed and in synchrony with one other, to create a good high- quality AC signal. These electro-mechanical interactions have traditionally underpinned the stability of the system. As such there are direct relationships between the mechanical characteristics of turbines and electrical voltage and frequency. In this section, we will explain the classical definition of stability and how that is changing with new technology entering the system. Kundur et al., (2004) provide the classic definition of the elements required for power system stability. They are: (1) frequency stability – the ability of a power system to keep frequency within reasonable bounds; (2) voltage stability – the ability to maintain steady voltages at all buses in the system; and (3) angular stability – the ability of resources in a system to maintain synchronism with the grid. While frequency is a system-level parameter, voltage and angular stability are more localized parameters. Frequency and voltage are well explained and understood in the power system literature. See Kroposki et al. (2017) and Milano et al. (2018) for excellent overviews of frequency and voltage concepts in a power grid. Angular stability can be explained by analogy. Imagine two runners running one behind the other at the same speed, tethered by a rope. The rope needs to be pretty taut but can tolerate a little bit of slack. There is a natural harmony between the two. If the runners want to accelerate or decelerate, the rope 5 IBRs, such as photovoltaic, wind power and batteries, are resources that interface with the grid through inverters, rather than directly through the turbine shaft. 7 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
provides a natural impetus – in other words, one runner can pull the other along to the new speed, as long as this is slow and controlled enough for the other one to respond. Instability arises when either runner accelerates or decelerates too quickly – either there will be too much slack, or a runner will be jolted forward/backward. This can be triggered by events internal to the system (one or both runners fall out of synchrony) or external (a dog runs into the runner’s path). 6 For a turbine generator to generate electricity there is a small time delay between the rotation of the mechanical turbine and the electrical signal that interfaces with the grid. One is always slightly out of phase with the other (as depicted in Figure 1). This is known as the rotor or phase angle. In a synchronized system, the differences between the phase angles of the generators across the system must be kept within operational limits. Instability can result in a generator suffering damage or disconnecting from the system. Figure 1: Phase angle Source: adapted from Kundur (1994) New technologies interface with the grid in a fundamentally different way from traditional generation sources. Traditional generation technologies are synchronous, which means that all mechanical rotating turbines are directly connected and synchronized with the grid, rotating with the same speed (for a frequency of 50 Hz this is 3000 rotations per minutes). By contrast, most IBRs are asynchronous because they have no rotating mass and connect to the grid via power electronic inverters, which convert DC electricity into grid-compatible AC electricity (Kroposki et al., 2017). 6 See also Basler and Schaefer (2005) for a technical overview of power system stability concepts. 8 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
Figure 2: Grid-following interface Source: authors The first generation of IBR use an industry standard control system, known as the phase-locked loop (PLL). As an IBR has no natural spinning mass, the PLL measures the phase of the voltage signal from the grid and then injects an AC voltage signal that matches the phase of the grid. This ensures that the signals are synchronized – they follow the grid signal (‘grid-following’ inverters) (Figure 2). While PLL-based inverters can have a range of alternative control schemes that trigger under certain conditions (for example, to provide frequency or voltage support), they continue to rely upon stable measurements of voltage from the grid. At low levels of IBR penetration, this approach has worked soundly. At higher levels of IBR penetration (30–40 per cent are often seen as trigger points) this introduces new categories of security issues that are not seen in traditional grids and lowers the ability of IBR to provide voltage and frequency support to the grid (AEMO, 2019; Bakke et al., 2019).7 As long as we continue to have synchronous resources in the grid, classical components of stability will continue to have relevance. However, as grids continue to transition to an inverter-dominated model, it is evident that the characterization of stability is itself evolving to reflect the additional considerations relevant to IBR and its control systems (Farrokhabadi et al., 2020). For the purposes of this paper, we have set aside a broad category of resource stability – which includes stability associated with synchronous generation (such as angular stability), as well as stability issues emerging as a result of higher IBR penetration. 2.3 Technology transition and integration The security challenges associated with the grid integration of new technologies can be broadly categorized into the impacts of (i) variability and uncertainty and (ii) new technology interfaces between generator and grid. While uncertainty has always been a factor in electricity market design since its inception (Schweppe et al., 1988), the nature of the uncertainty has changed. Power system operators must manage for the uncertainties introduced by more intermittent forms of generation, such as wind and solar, which for the purposes of security relate more to short timescales (minute or second timescales) and can impact upon system parameters such as frequency and voltage. These can include, for example, changes in wind or solar generation due to weather patterns or clouding. It can also include risks of greater correlation or co-movement between sets of resources. 7 These include classic voltage stability issues, but also new interactions between the grid and IBR control systems, and between IBRs – including previously unobserved interactions where IBRs oscillate with one another (Rehman et al., 2019; Wang and Blaabjerg, 2019). 9 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
Second, due to the link that existed between turbine mechanics and electric networks, there have been important physical characteristics of synchronous resources, which have traditionally supported the security of the system but are less available in the transition era. These include: • Inertia: the spinning mass of rotating turbines have a store of kinetic energy, which is naturally and instantaneously released into the system if there is imbalance in demand and supply. Inertia naturally slows the rate at which system parameters (such as frequency and voltage) change in response to disturbances. This slows the rate of degradation of the parameters but also their recovery. • Fault current: On the occurrence of a fault, synchronous resources will naturally and instantaneously inject high levels of current into the fault. This assists with limiting the rate at which voltage degrades during fault conditions. In other words, it keeps voltages ‘stiff’. • Synchronization effect: There is a natural synchronization effect between all generators connected within a network (i.e. to use the walking analogy used earlier, the rope that ties all participants together). The electrical response that underpins these characteristics are inherent to the traditional synchronous generation technology and are a natural response to a power system disturbance. Operational control for AC grids has heretofore relied upon these characteristics to underpin system security (Milano et al., 2018). With the interface between generator and grid increasingly moving towards power electronic converters, these existing services are becoming scarce and result in weaker grids that are less resilient to disturbances. Inverter control and systems technology are advancing at a rapid rate to meet the challenges of system security. IBRs are already capable of delivering rapid and precise frequency and voltage response services, by injecting or withdrawing active or reactive power into the system. This has been already implemented in some markets and is expected to be implemented in many others. On a more fundamental level, there is also the development of new modes of control from IBRs that aim to resolve some of the interface issues between first-generation IBRs and the network. The approaches taken to meet the challenge vary from fitting new technology into existing modes of control, to creating entirely new operational paradigms more suited to an inverter dominated grid (Matevosyan et al., 2019). Research and development have been underway on classes of ‘grid-forming inverters’,8 which adopt new control schemes that have less reliance on grid measurements. There are many variations on ‘grid-forming’ control from those that try to mimic the full response of synchronous machines (virtual synchronous machines), while others adopt alternative control strategies (for example, virtual oscillators). However, it is still early days, and few have been piloted as of yet. 9 What does this all mean for policy and market design? It is clear that electricity market design needs to be flexible and adaptable to reflect these ongoing changes in technology. This will inevitably involve trade-offs between maintaining existing control modes and using new models and technologies. Immediate risks to security may require a near-term reliance upon what operators know works. However, testing and piloting of new technologies on an ongoing basis is important to ensure that, once proven, new technologies can participate in the market without prejudice. An ideal approach is one that 8 The current generation of inverters operate as so-called ‘grid-following’ sources, meaning that they track the phase angle of the grid in order to synchronise their output. Grid following inverters are best viewed as a source of current, and are less able to assist in keeping voltages stiff at the connection point in the event of disturbances (such as faults). By contrast, the control schemes of grid forming inverters act as a voltage source, which allows them to contribute to keeping voltages stiff at the connection point. 9 There has been some piloting of grid-forming technologies on large-scale grids, including the Dalrymple ESCRI battery project in South Australia (https://www.escri-sa.com.au/). In many cases, these have focussed upon microgrids or smaller-scale islanded grids (Schomann, 2019). 10 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
effectively bridges the gap between the economics of service delivery and the engineering of resource control, though pragmatism may require operators and policymakers to favour one over the other at different times. The roles and incentives of decision makers and regulators in this domain must also be considered. 2.4 Power system requirements Table 1 outlines the physical requirements for a secure power system and highlights the scarcities and phenomena emerging as a result of the transition away from legacy synchronous generation towards resources that are inverter-connected and variable in nature. In the following sections, we will draw out the essential elements of frequency and voltage control that are important for economic characterization. However, a detailed treatment of engineering control applications for frequency and voltage under high IBR penetration is out of the scope of this paper. Interested readers are instead referred to the following resources: CIGRE, 2016; EPRI, 2019; Milano et al., 2018; Tielens and Van Hertem, 2016. Table 1: Summary of technical power system requirements Parameter Physical requirements Emerging scarcities and concerns type Inertia and primary response • Declining inertia affecting frequency response Frequency stability • Reduced synchronous load affecting frequency Secondary response System damping • More rapid frequency variation within periods Tertiary response • Voltage induced frequency dips (VFID) Static voltage control • Declining fault levels affecting voltage stability stability • Voltage Local Potential for increased voltage oscillation Dynamic voltage control • Voltage variations with intermittent resources • Voltage waveform distortions from IBR System strength and fault levels Local • Reduced inertia affecting voltage recovery Angular stability • Declining inertia affecting angular stability Resource stability • Reduced synchronization effect Resonance stability Local • Harmonic interactions between IBR units • Delayed active power recovery in IBR Control systems stability dominated grids Source: compiled from Kroposki et al. (2017); Milano et al. (2018); Shair et al. (2019); CIGRE (2016); AEMO (2018) 2.4.1 Frequency management and inertia The incumbent approach to frequency management in response to a system event or contingency involves three stages: primary response (PFR) to arrest large frequency deviations and recover it to initially acceptable levels; secondary response (SFR) to regulate smaller deviations within normal operating levels; tertiary response (TFR) seeks to replace reserves that used to provide faster responses, to provide some measure of cushion. 10 Some markets, such as the National Electricity Market of Australia (NEM), dispense with TFR reserves and instead rely upon the redispatch of the power system to keep the new balance in the power system. If frequency degrades past normal 10 As we outline below, a shift towards 100% IBR may entail new control modes. In the near term, inertia will continue to be a relevant parameter but may degrade in importance over time. 11 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
operating bounds, emergency control schemes (such as under-frequency load shedding) trigger to attempt to avoid cascading failure. In each case, frequency is controlled through injecting or withdrawing active power from the system. Figure 3: Frequency response under contingency Source: authors, adapted from Mancarella et al (2017) The impacts of lower inertia on traditional frequency control have been well studied in recent years (Milano et al., 2018). Inertia reduces the instantaneous rate at which frequency changes (known as the rate of change of frequency, or ROCOF). It also affects how low the frequency gets (the nadir).11 It also slows down the recovery of the system (Figure 3). Frequency is a critical parameter for existing turbine generation and the network, and there are limits to the ROCOF and nadir that this equipment can handle. As such, the grid relies upon these measures being between viable operating bounds, or else risking cascading failure or equipment damage. There are a range of operational levers and responses to deal with the loss of inertia, as a result of the retirement or reduced merit order dispatch of synchronous generation. First, and most obvious, would be to increase the level of inertia in the system. This can be done through re-scheduling synchronous generation (SG), by installing synchronous condensers (‘syncons’) to replace lost or retired SG, or to modify SG to allow operation in a synchronous condenser mode. ‘Primary response, especially from fast-acting resources, can be used to substitute for and be co-optimised with inertia to control the nadir (Mancarella et al., 2017). This can be sourced either as reserve quantities based on triggers or through active control schemes, such as droop-based frequency control mechanisms (PFC). There is also active research and piloting of IBR control schemes that seek to mimic synchronous resources (so-called synthetic inertia or virtual synchronous machines) to provide as-close-to instantaneous inertial response’ (Milano et al., 2018). 11 In this section we mainly talk about an under-frequency event, for example in respect of a loss of a generation unit. Frequency operating frameworks must also cater to over-frequency events, such as the loss of load, and the same principles apply. 12 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
Figure 4: Constance frequency operating modes Source: Ramasubramanian (2018) A further opportunity lies in controlling the quantum of risk in the system. The amount of inertia or primary response is sized to cater for a particular risk level – usually the largest generator in the system. Thus control over the sizing of the contingency, by controlling the dispatch of the largest generator, has been argued to serve as a partial substitute (Badesa et al., 2018; Nedd et al., 2018; Pϋschel- Lovengreen and Mancarella, 2018, Mancarella et al., 2017). This latter approach assumes that the inherent risk in the system is correctly captured by the sizing metrics, which is appropriate for primary control that focuses upon unit contingency events. It has less application in respect of frequency dynamics that are affected by risks other than unit contingencies – for example in relation to generation variability. The role of system thresholds is also important. The operational limits of the system (for example thresholds for ROCOF and nadir) are guided by the tolerance capabilities of the system within it. Synchronous machines that rely on granular control of large machinery have different tolerances to electrically coupled resources that convert from DC. Higher technical requirements can also impact the amount of inertia and frequency response that is needed.12 While the policy response to frequency challenges has focussed upon mitigating the impacts of technology changes to frequency sensitivity, it is also possible that the transition to an inverter-based grid requires new operational paradigms and ways of running the system. In a system with 100 per cent inverter-based generation, frequency may be a less relevant parameter. Different regimes may be required that continue to maintain power balance without the use of frequency as a control variable. Constant frequency operation is a proposed mode of operation where IBRs aim to maintain a constant frequency, rather than a bounded range, through controlled injections of active power (Figure 4). For an overview, see Ramasubramanian, (2018). For more detailed analysis, see Ramasubramanian et al (2018). 12 EirGrid has a programme of work to transition grid code requirements for ROCOF from 0.5 Hz per second to 1 Hz per second (Eirgrid and SONI, 2018). National Grid in the UK is also undertaking changes to its distribution code relating to ROCOF relay settings (National Grid, 2020b). 13 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
2.4.2 Voltage stability and system strength in weak grids Voltage is a local parameter and affects the operation of both synchronous resources and IBRs. In general, the relationship between voltage and power generation at a node is as per Figure 5. Voltage at every node needs to be operated above a critical point. Operating closer to the critical point makes the voltage more sensitive to changes in power flows, and more at risk of instability and collapse. As such, voltages are affected not just by the steady state operating point, but also by the variability of power flows. The addition of varying generation and load, such as from renewables, can thus impact upon the stability margin of voltage in certain locations around the grid (Pierrou and Wang, 2019). Figure 5: Voltage stability margins Source: authors Voltage control is achieved through managing the reactive power flows around the system (as shown in Figure 6). This is typically achieved through a combination of network control schemes and equipment at the node or regional level, or by changing the set-points of generators at different locations. Control schemes at the generator level can also assist with voltage stability. Figure 6: Effects of reactive power on voltage Source: authors In addition to traditional voltage stability, an emerging concept is that of ‘system strength’. A weak grid is one where the voltage is sensitive to changes in active power or reactive power (CIGRE, 2016; Dozein 14 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
et al., 2018). By contrast, a strong grid is one where the voltages are ‘stiff’ and exhibit much less sensitivity to power changes and to system events. There are physical characteristics of synchronous generators (and other resources, such as synchronous condensers) that have traditionally supported system strength. By contrast, IBRs have characteristics that have limited their ability to contribute to system strength (Gu et al., 2019a, 2019b; Shair et al., 2019). First, IBRs are limited in the amount of current they can inject during a fault. Second, and perhaps more fundamentally, the first generation of IBR technology (using PLLs protocol) themselves rely upon a ‘stiff voltage’ to synchronize with the grid. A large presence of online synchronous generators inherently slows the dynamic changes in the system, thereby allowing grid-following inverters to accurately track the grid voltage (Matevosyan et al., 2019). A weak grid is susceptible to a number of issues including: (1) mal-operation of fault-protection systems; (2) mal-operational of inverter control systems; (3) voltage oscillations following faults; (4) angular instability for synchronous generators; and (5) sub-synchronous control interactions between IBRs. With the retirement of synchronous fossil fuel generators, weak grid conditions are increasingly being observed, especially in areas of high renewable generation. System strength is a complex issue and represents an interaction of many different factors. The inherent responses of synchronous resources, such as fault levels and synchronising power are important, as is the specific control systems employed by IBRs. System strength is also a very locational issue and depends upon many factors, including the combination and location of generators online, and the transmission configuration. There are some high-level metrics that provide high-level screening tools as to whether a system is strong (NERC, 2017; Wu et al., 2018). Often however, system operators need to undertake detailed and computationally intensive simulations to determine whether particular combinations of resources will give grid voltages that are sufficiently resistant to disturbances (Matevosyan et al., 2019). In resolving near-term system strength issues, operators have looked towards re-scheduling synchronous generation, or installing synchronous condensers (Jia et al., 2018; Kenyon et al., 2020; Richard et al., 2019). In other cases, the tuning of inverter settings and constraints on IBRs has been utilized as alternative measures (Filatoff, 2020). Longer term, newer technology and control schemes (such as grid-forming inverters) are being considered (Matevosyan et al., 2019). 3. The economic characterization of system security 3.1 The system services value chain: elementary services, system products and power system requirements In the sections above, we have described the technical challenges associated with power system security. In this section and hereafter, we begin to frame some of the economic characteristics of system security. As a start, we think it is a helpful analogy in this situation to think about system security as a value chain. This allows us to understand what the ‘raw materials’ of power system security are, how they are packaged or segmented into different products or types of responses, and ultimately how those products combine in a manner that fulfils the needs of the users (see Figure 7). What are the raw materials of power system security? The requirements for the control of frequency and voltage can be reduced at a fundamental level to four elementary services (Rebours, 2008). These are specified as: • upward active power – increasing injection or decreasing withdrawal of active power • downward active power – decreasing injection or increasing withdrawal of active power • upward reactive power – increasing injection or decreasing withdrawal of reactive power • downward reactive power – decreasing injection or increasing withdrawal of reactive power. 15 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
While the characterisation of these elementary services as it relates to concepts such as active power (frequency) or voltage (reactive) reserves is intuitive, this concept can also be extended to concepts such as inertia and system strength. As outlined in Section 2, inertia is stored energy released during disturbances. Fault levels relate to the natural injection of current (and hence power, either reactive or active) in response to system faults. More complex system strength phenomena also relate to the ways in which voltage-sources and current-sources inject or withdraw power in a network given either through control schemes or inherent physical responses. A common approach then, in many market designs, is to designate a particular type of response, in other words, how active or reactive power is to be injected or withdrawn in the system. This has involved the creation of products based on a variety of factors, including: (i) what the triggering event is (if any); (ii) the quantity and quality of the response; (iii) the speed and duration of the response and (iv) notice times. Providers of these products must have the headroom or reserve to deliver these products, over and above what they may be delivering as part of energy dispatch. These definitions are important determinants of efficiency and liquidity in these markets, which in turn influence system security (Oren, 2001). There is an inherent trade-off in the design of system service markets between homogenization of products to ensure competition, and product differentiation to meet system needs and optimize quality of delivery. Market designers have to decide upon the optimal balance between too many products and too few products (Pollitt and Anaya, 2019). Works such as Badesa et al (2020) and Greve et al (2018) suggest that artificial product boundaries can in some cases be replaced where all providers bid their technical characteristics (in addition to price and quantity), thereby allowing a market operator or co-ordinator to make optimal trade-offs between them. However, there is also a distinction between those products which have a controlled actuation (i.e. via centralized or decentralized controllers) and inherent actuation (i.e. on the basis of the physical characteristics of the generator or resource). System strength and inertia from synchronous generators fall into the latter category. While control-based responses can be switched on or off, inherent responses are naturally provided when the resource is online. They are binary in nature and cannot be scaled up or down. Figure 7: System security value chain – elementary services, system security products and requirements Source: authors Recent security outcomes have underscored the importance of inherent responses (inertia, system strength and synchronization effect) for system security. These inherent responses have unique 16 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
aspects which may create complexities for the design of economic products around these responses. These include (i) definability of the services, (ii) measurability and modelling of system impacts, (iii) unit commitment, and (iv) the inability to separate the delivery of different services. Definability. Power system security in an asynchronous grid is very much an emerging area of research, and new dynamic behaviors are being observed every day. This has made the definition of power system requirements much more difficult, especially where a variety of factors interact in different ways. System strength is one such requirement. It is not clear if is it a power system requirement or a product. Also, we do not know whether a response type for individual providers can be defined in a common, consistent and effective way. Measurability and modelling. There is also the challenge of measuring and modelling dynamic interactions. This is important for product quantification and for understanding trade-offs between the quantity of the product and other power system factors. For example, deriving constraint equations under security constrained unit commitment and dispatch (SCUC/SCED) market designs requires a clear mathematical relationship between variables such as the online status of synchronous units and the dispatch (in MW) of generators in the network. This means that there needs to be clear metrics for the security condition you are constraining the system for. While this may be carried out for some of the new system requirements such as for co-optimization of inertia and frequency response (Mancarella et al., 2017) it is challenging for other emerging security phenomena, such as low system strength in areas with plenty of renewables. In fact, while there are some standardized metrics for system strength (including a range of metrics that are modifications or adaptations of the short circuit ratio or SCR), there is a question of whether these metrics adequately capture the phenomenon. Although they may be sufficient for planning and connection processes, it is not clear if they are adequate for dispatch and scheduling. On the other hand, detailed transient simulations provide a better and more accurate understanding of whether the system is resilient to certain events or contingencies (such as faults). However, they are computationally intensive and sensitive to many factors. In markets such as the NEM where system strength is already a scarcity, operators have resorted to creating a schedule of generator configurations that provide comfort around system strength sufficiency (Ela et al., 2019). Ultimately, a set of viable parameters suitable for commitment and dispatch is required and should emerge as more is understood about the phenomena. Unit commitment. Inherent response products are typically binary responses, in other words, they are based on whether the unit is synchronized to the system. The unit provides its full response if it is synchronized, and zero if it is not. These responses do not scale up or down with the level of generation, which is important for approaches to commitment and dispatch. Unit commitment creates non- convexities in dispatch optimization,13 which makes locational marginal pricing less meaningful. Some designs have adopted linear relaxations of the integer products to obtain meaningful pricing of inertia (Badesa et al., 2020). In other markets, the challenges of revenue inadequacy flowing from centralized unit commitment have relied upon ‘uplift and clawback’ side-payment frameworks, and upon alternative price formations for energy (such as convex-hull approximations) to minimize those uplift payments (Gribik et al., 2007). However, it is questionable whether energy price formation is an appropriate tool when the services being delivered are not always linked to the delivery of active power (for example, a synchronous condenser). Separability. Particularly for inherent response products, there is an inability to separate the delivery of different services. For example, if online, a synchronous generator will provide both inertial, fault level and system strength contributions. This is of relevance when the products have fundamentally different economic characteristics. Simultaneous delivery must also be reflected in how these products are priced. 13 With a convex optimization, there can be only one optimal solution, which is globally optimal, whereas non-convexity causes multiple feasible regions and multiple locally optimal points within each region. 17 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
3.2 An economic characterization of system security and services The electricity system is not unique in its need to manage security and quality of service (Pollitt and Anaya, 2019). Economic frameworks should guide how the comprehensive needs of the power system are met. Electricity and other essential infrastructure goods can be usefully assessed through the perspective of public good theory. In this section, we draw upon the rich body of literature in public good theory to appropriately characterize system security and the products required to deliver system services. 3.2.1 The nature of goods The modern identification of a public good, and its distinction from a private good, is based on characteristics of excludability and rivalry (Table 2). A pure public good is one that clearly exhibits properties of (i) non-excludability and (ii) non-rivalry, while a private good is clearly excludable and rival. A good is excludable if the users of the good can be easily excluded from the enjoyment of the good (where the marginal cost of exclusion is low). Rivalry pertains to jointness of use. A rival good is one where the consumption of the good by one person diminishes the ability of others simultaneously to consume the good – where the marginal cost of extending the use of the good to a new user is high (the marginal cost of extension). Non-rival goods can be enjoyed by multiple users without diminishing its value. Table 2: Classification criteria for different types of goods Rivalry Non-rival Rival Excludable Club goods Pure private goods Excludability Non-excludable Pure public goods Common pool resources (CPR) Source: Ostrom (1990) Seminal works by Ostrom (2003, 1990, 1977), Buchanan (1965) and others identified additional classifications that do not neatly fall into the classification of a pure private or public good. Ostrom’s Nobel Prize winning work developed the notion of a common pool resource (CPR) – a good that is rival but not excludable. Furthermore, goods that are not rival but are excludable are termed as a club or toll goods. CPRs are subject to the tragedy of the commons. This is a phenomenon where individual users acting in their own interests are incentivized to exploit a resource. Uncontrolled, this can lead to overexploitation of the resource to the detriment of the entire user group. 14 This is particularly relevant for aspects of electricity service as it relates to the utilization rates of resources (static efficiency) and the incentive to invest in new capacity (dynamic efficiency) (Kiesling, 2008). Few, if any, goods are perfectly non-rival. At some thresholds of supply, congestion begins to occur and one person’s use of a good begins to impact its enjoyment by others. There is a category of goods known as congestible common pool resources, which initially behave like public goods when there is excess capacity. However, as more users are added, congestion starts to occur and the good becomes increasingly rival. A common example of this is a toll road/highway which is initially non-rival when there is excess capacity, but congestion effects and rivalry will begin to appear as the number of users increase. The concept of CPRs has been previously applied been applied to different aspects of the 14 A common example of the tragedy of the commons is environmental air quality or emissions, where the marginal cost of emissions on an individual is very low and there is no direct incentive not to emit. If all individuals act in their interests the cost on society as a whole is nevertheless significant. 18 The contents of this paper are the authors’ sole responsibility. They do not necessarily represent the views of the Oxford Institute for Energy Studies or any of its Members.
You can also read